The bug in our code is that pari does not raise an error, it dies once you
run out of memory but thats not a good thing to wait for.
On Monday, January 21, 2013 4:36:03 PM UTC, Javier López Peña wrote:
>
> On Monday, January 21, 2013 3:28:29 PM UTC, Simon King wrote:
>
>> OK, then we should at
On Monday, January 21, 2013 3:28:29 PM UTC, Simon King wrote:
> OK, then we should at least fix the documentation of LLL_gram.
>
Our LLL_gram does this:
if self._nrows != self._ncols:
raise ArithmeticError("self must be a square matrix")
n = self.nrows()
# maybe sho
Hi,
On 2013-01-21, Volker Braun wrote:
> --=_Part_128_10880395.1358779116647
> Content-Type: text/plain; charset=ISO-8859-1
>
> Input really must be positive definite, not just definite with any sign.
OK, then we should at least fix the documentation of LLL_gram.
Cheers,
Simon
--
You rece
Default should always be to check the input over calling library routines
with potentially undefined behavior.
On Monday, January 21, 2013 2:37:15 PM UTC, Simon King wrote:
>
> Problem: If the "check" option is True by default then the user will
> have to wait for the positive definiteness test
Input really must be positive definite, not just definite with any sign.
Function: qflllgram
Section: linear_algebra
C-Name: qflllgram0
Prototype: GD0,L,
Help: qflllgram(G,{flag=0}): LLL reduction of the lattice whose gram matrix
is G (gives the unimodular transformation matrix). flag is optiona
Hi Volker,
On 2013-01-21, Volker Braun wrote:
> LLL_gram() is a good deal faster than checking positive definiteness:
Sorry, your post came while I wrote my post. If checking positive
definiteness can be a bottle neck then my suggestion to test it by
default is not good.
> I guess this means th
Hi y'all,
On 2013-01-21, Volker Braun wrote:
> This uses Pari lllgramint(), which assumes that the matrix is positive
> definite. If the matrix is not positive definite, Pari may not return.
>
> sage: D.is_positive_definite()
> False
Then we have at least a bug in the documentation. Namely, it
LLL_gram() is a good deal faster than checking positive definiteness:
sage: m = random_matrix(ZZ,20)
sage: %time (m*m.transpose()).LLL_gram()
CPU times: user 3.92 s, sys: 0.00 s, total: 3.92 s
Wall time: 3.94 s
200 x 200 dense matrix over Integer Ring
sage: %time (m*m.transpose()).is_positive_de
I stand corrected. Note however that lllgramint = qflllgram(-, 1) is
suggested by pari itself:
? lllgramint(D)
*** at top-level: lllgramint(D)
*** ^-
*** not a function in function call
A function with that name existed in GP-1.39.15; to run in backward
com
Its not a bug, its undefined behavior for invalid input.
Also, I don't think we should ever use lllgramint = qflllgram(-, 1). Thats
the toy implementation, you want the adaptive floating point implementation
(flag=0) for real work.
http://permalink.gmane.org/gmane.comp.mathematics.pari.devel/2
Hi Volker,
I think lllgramint() is deprecated, we should call gflllgram(D,1) instead.
The bug remains the same though.
Cheers,
J
On Monday, January 21, 2013 1:41:35 PM UTC, Volker Braun wrote:
>
>
>
> On Wednesday, December 19, 2012 11:07:03 PM UTC, William wrote:
>>
>> > sage: D=Matrix(IntegerM
Hi Simon,
it seems to be a bug in Pari, the whole computation ends up calling
gflllgram(D._pari_(),1)
which doesn't end.
Constructing the same matrix directly in pari and calling the function
there also doesn't end.
I have tried changing the values for the flag and same behavior happens
with
On Wednesday, December 19, 2012 11:07:03 PM UTC, William wrote:
>
> > sage: D=Matrix(IntegerModRing(),
> [[-1,1,0,1,1,0],[1,-3,1,0,0,0],[0,1,-2,0,0,0],[1,0,0,-3,0,0],[1,0,0,0,-4,1],[0,0,0,0,1,-5]]);D
>
> > [-1 1 0 1 1 0]
> > [ 1 -3 1 0 0 0]
> > [ 0 1 -2 0 0 0]
> > [ 1 0 0 -3 0 0
Hi Sage supporters,
bump!
Is there really no answer to Ian Hambleton's question that he asked a month
ago?
Can it really be so difficult to do LLL_gram on a 6x6 integer matrix?
Cheers,
Simon
Am Donnerstag, 20. Dezember 2012 00:07:03 UTC+1 schrieb William:
>
> I'm forwarding this to sage-suppo
I'm forwarding this to sage-support...
On Dec 19, 2012 8:52 AM, "Ian Hambleton" wrote:
>
> Dear Professor Stein,
> I am trying to use the Sage diagonalization routine found in:
>
>
http://www.sagemath.org/doc/reference/sage/matrix/matrix_integer_dense.html
>
> The example given in the Sage docume
15 matches
Mail list logo
